> For the complete documentation index, see [llms.txt](https://docs.walnutai.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.walnutai.ai/development-and-infrastructure/walnutai-coder/dev-and-qa-workflow-overview.md).

# Dev & QA Workflow Overview

WalnutAI unifies development and quality assurance into a single, seamless workflow. In Dev Mode, it guides you through planning, understanding, and building features-whether for new projects or existing applications. In QA Mode, it ensures reliability through test design, automated test generation, and continuous stability. The agent serves as the default interface, intelligently switching between modes based on the task to deliver a smooth, end-to-end development experience.

#### Dev Mode

In Dev Mode, the agent assists with development through the following modes:

* **Plan Mode**: Acts as the architect, transforming high-level prompts into a structured, step-by-step technical roadmap before any code is written.&#x20;
* **Ask Mode**: Enables users to clarify requirements, gather insights, and receive guidance throughout the development process.

Additionally, the **Tasks** section in Dev Mode provides visibility into all tasks assigned to the user. By selecting a task, users can directly begin working on and resolving it.

<figure><img src="/files/rXAjcDsEa1zO1CFE5zKK" alt=""><figcaption></figcaption></figure>

#### QA Mode

In QA Mode, the agent supports testing and quality assurance through the following modes:

* **Test Planner Mode**: Acts as the strategist, identifying critical paths that require testing and defining the overall validation strategy.
* **Test Generator Mode**: Functions as the QA engineer, automatically generating executable unit, integration, or end-to-end tests.
* **Test Healer Mode**: Acts as the maintainer, executing tests, diagnosing failures, and automatically updating code or test logic to restore stability.

<figure><img src="/files/U2nRbwl7G6xJdbs5Ybnu" alt=""><figcaption></figcaption></figure>

* **Repository & Custom Method Workflow:** Within QA Mode, users can manage repositories and create custom methods. Before proceeding, ensure that the selected project is properly configured with the appropriate repository.

  * **Open Repo**: Clicking **Open Repo** allows you to select a local folder, where the repository will be cloned.

  <figure><img src="/files/TERyahHLPqiBVQC9IY6w" alt=""><figcaption></figcaption></figure>

  * **Initialize:** Sets up the repository and creates a sample method.

  <figure><img src="/files/0v9QzLY9itkzYp1wW2YU" alt=""><figcaption></figcaption></figure>

  * **Custom Method Creation**: Selecting “Custom Method” initiates a series of structured questions to capture the method’s purpose, required actions, and inputs. Based on these inputs, the system automatically generates the requested method.

  <figure><img src="/files/ELTu4ZQ6DCf3myuRhNuW" alt=""><figcaption></figcaption></figure>

  * **Sync**: The created method is synced to WalnutAI. Once synchronized, it can be accessed via Admin Settings under Action Settings or within test steps using “/ → Actions.”

  <figure><img src="/files/cDaEq7z4sGeABpAC4oGQ" alt=""><figcaption></figcaption></figure>

  * **Version Control**:
    * **Push**: Upload the created method to the repository.
    * **Pull**: Fetch the latest updates from the repository.
  * **Debug Mode**:

    * Activate Debug Mode to test and troubleshoot custom methods.

    <figure><img src="/files/eWdr4b7azSPo3CUJpVaB" alt=""><figcaption></figcaption></figure>
  * To debug, create a test case, link the custom method, and execute the test case to validate functionality.

  <figure><img src="/files/X0LAAaLRVEQ0bl5jvVtR" alt=""><figcaption></figcaption></figure>

**Human-in-the-Loop Workflow**

To ensure you maintain total control, the agent operates within a strict safety and collaboration framework:

* **Interactive Q\&A Mechanism:** If a prompt is ambiguous (e.g., "Add login"), the Agent pauses to ask targeted questions about authentication methods or database preferences rather than making assumptions.

<figure><img src="/files/pfdhYQg9cujFw6Pe04rH" alt=""><figcaption></figcaption></figure>

* **Approval Framework**: The Agent must receive your explicit permission before performing impactful actions, including:

  * Running terminal commands.
  * Deleting or restructuring project files.
  * Applying large-scale refactors or database migrations.

  <figure><img src="/files/t9EuCzJxeGCdXYeUWzJJ" alt=""><figcaption></figcaption></figure>
